1. Getting Biologists off ACID Ryan Verdon 3/13/12

2. Outline Thesis Idea Specific database Effects of losing ACID What is a NoSQL database Types of NoSQL databases with examples

3. Thesis Idea Amount of biological data is growing exponentially Hardware cannot keep up with the growth

4. Database Survey In 2003, a survey was conducted of 111 biological databases. 96 (i.e. 87%) of the 111 considered databases have Hypertext references to other databases, 40 to 44 (i.e. 36% to 40%) are implemented as flat files, 41 (or 42) (i.e. 37%) are implemented using a relational database management system, 7 (i.e. 6%) use an object database management system, 3 (.i.e. 3%) use an object-relational database management system, and all databases collect data from different sources.

5. Specific Database

6. Flybase Primary biology database on the insect family Drosophilidae (fruit fly) Important research from fruit flies includes – Genes – Recombination – Signaling networks (important for major diseases) – Stem cells – Growth control

7. Info on the Flybase database Single server relational database Over 135 tables Uses Chado schema Gives data a “type” Creates a graph that relates the different types

8. Effects of losing ACID

9. ACID Atomicity Consistency Isolation Durability

10. CAP Theorem Consistency Availability Partition tolerance

11. BASE consistency model Basically Available Soft state Eventually consistent Given enough time where no changes are made, all replicas will see the same data

12. NoSQL

13. What is a NoSQL database Any non-relational database – Hierarchical – Graph – Object oriented – Etc.

14. Why were NoSQL databases created To overcome limitations of relational databases – Predefined layout – Scaling – SQL – Large feature set

15. Major types of NoSQL databases Key-value stores Column-oriented databases Document based stores

16. Key-value stores Stored values are indexed for retrieval by keys Can store unstructured or structured data Similar to DHTs

17. Example: Dynamo Created by Amazon, only used internally Highly available even in the face of continual failures Amazon realized many services only need primary-key access – Best seller lists – Shopping carts – Session management

18. Column-oriented databases Contain extendable columns of closely related data Can greatly benefit from compression

19. Example: HBase Apache open-source project Based off of Google’s Bigtable database Ties in closely with Hadoop and map-reduce

20. Document based stores Data stored as a collection of documents Documents can have any number of fields and any length Documents are accessed via a unique key Capabilities of the query language heavily depend on the implementation

21. Example: MongoDB Started by 10gen Supports – Replication – Map-reduce – Sharding Used by lots of large companies including – Disney – Craigslist

22. Questions ?